ZTE UMTS HSDPA Packet Scheduling Feature Guide

HSDPA Packet Scheduling WCDMA RAN

Feature Guide

Operator Logo

HSDPA Packet Scheduling Feature Guide

ZTE Confidential Proprietary © 2010 ZTE Corporation. All rights reserved. I


Version Date Author Approved By Remarks

V4.5 2010-10-15 Wang Yue Peng Bei ,Hu Ye,

© 2010 ZTE Corporation. All rights reserved.

ZTE CONFIDENTIAL: This document contains proprietary information of ZTE and is not to be disclosed or used without the prior written permission of ZTE.

Due to update and improvement of ZTE products and technologies, information in this document

is subjected to change without notice.


ZTE Confidential Proprietary © 2010 ZTE Corporation. All rights reserved. II

TABLE OF CONTENTS

1 Functional Attribute ............................................................................................ 1

2 Overview .............................................................................................................. 1

2.1 HSDPA Fast Scheduling....................................................................................... 2

2.2 HSDPA Flow Control ............................................................................................ 3

3 Technical Description......................................................................................... 3

3.1 HSDPA Resource Allocation Scheme .................................................................. 3

3.2 HSDPA Scheduling Algorithms ............................................................................ 4

3.2.1 MAX C/I Algorithm ................................................................................................ 5

3.2.2 RR Algorithm ......................................................................................................... 5

3.2.3 PF Algorithm ......................................................................................................... 5

3.2.4 Summary of Scheduling Algorithms ..................................................................... 7

3.3 HSDPA TFRC Selection Algorithm....................................................................... 8

3.3.1 HS-SCCH Code and Power Selection ................................................................. 8

3.3.2 TBSIZE, Number of HS-PDSCH Channelization Codes, Modulation and Power Selection.................................................................................................... 9

3.4 HSDPA Flow Control Algorithm .......................................................................... 12

3.4.1 Flow Control Implementation Method................................................................. 12

3.4.2 Flow Control Algorithm........................................................................................ 13

3.5 Measure of HS-DSCH Required Power ............................................................. 16

3.6 Impact of HSPA+ on Scheduler.......................................................................... 17

3.6.1 Impact of introducing 64QAM Modulation Technology on Scheduler................ 17

3.6.2 Impact of introducing MIMO on Scheduler ......................................................... 20

3.6.3 Impact of introducing DC-HSDPA on Scheduler ................................................ 22

3.7 Dynamic power sharing in Multi-carrier .............................................................. 22

4 Parameter Description ..................................................................................... 24

4.1 HSDPA Scheduling Algorithm Parameters ........................................................ 24

4.1.1 Channel Quality Weight ...................................................................................... 24

4.1.2 SPI_0 SPI_1...SPI_15......................................................................................... 25

4.1.3 HS-PDSCH Total Downlink Power Allocation Method ....................................... 25

4.1.4 HS-PDSCH Measurement Power Offset ............................................................ 25

4.1.5 Support Type of RLC Flexible PDU Size Format ............................................... 26

4.1.6 Non-Conversational service Maximum MAC-d PDU Size Extended ................. 26

4.1.7 Conversational service Maximum MAC-d PDU Size Extended......................... 26

4.2 Dynamic Power Sharing in Multi-carrier Parameters ......................................... 27

4.2.1 Transmission Power ........................................................................................... 27

4.2.2 Power-sharing Ratio ........................................................................................... 27

5 Weight Mapping Table...................................................................................... 28

6 Glossary ............................................................................................................. 29


ZTE Confidential Proprietary © 2010 ZTE Corporation. All rights reserved. III

FIGURES

Figure 2-1 n of Node B HSDPA ............................................................................................... 2

Figure 3-1 Resource allocation during UE scheduling .......................................................... 10

Figure 3-2 Flow control procedure......................................................................................... 13

Figure 3-3 Leaky bucket flow control scheme ....................................................................... 14

Figure 3-4 Static power sharing in Multi-carrier .................................................................... 23

Figure 3-5 Downlink power after introducing HSDPA ........................................................... 23

TABLES

Table 3-1 HS-PDSCH start code Nos ................................................................................... 18

Table 4-1 Parameter settings in OMMB ................................................................................ 24

Table 4-2 Parameter settings in OMMR ................................................................................ 24

Table 4-3 OMMB configuration parameters filed .................................................................. 27

Table 5-1 CQI weight mapping table ..................................................................................... 28


ZTE Confidential Proprietary © 2010 ZTE Corporation. All rights reserved. 1

1 Functional Attribute

System version: [RNC V3.09, Node B V4.09, OMMR V3.09, OMMB V4.09]

Attribute: [Optional functions]

Involved NEs:

UE Node B RNC MSCS MGW SGSN GGSN HLR

√ √ √ - - - - -

Note:

*-: Not involved.

*√: Involved.

Dependency: [None]

Mutual exclusion: [None]

Remarks: [None]

2 Overview

After the High Speed Downlink Packet Access (HSDPA) technology is introduced to

WCDMA, the MAC-hs layer is added to both Node B and UE. The HSDPA of Node B

implements the following functions:

1 Receiving and storing HS-DSCH data frames from UE on lub interface.

2 Performing cell-based multi-UE fast scheduling.

3 The Adaptive Modulation and Coding (AMC) provides HS-PDSCH air interface

transmission format (Number of channelization codes, modulation mode, and TB

size).

4 Transmitting HS-SCCH control information and HS-DSCH data information.

5 Demodulating messages such as ACK, NACK and CQI carried on HS-DPCCH.

6 Performing downlink HS-DSCH flow control.

The following figure shows the functional implementation of MAC-hs:



Figure 2-1 n of Node B HSDPA

FP

AMC

Hsdpa

Scheduler

Harq

Entity

HSDPCCH

ACK/NACK/CQI

MACD PDU

Data PoolACK/NACK

CQI

Select Tbsize

Retrans

QueSelect Sch

UE/Que

HSSCCH Info

FP

AMC

Harq

Entity

HSDPCCH

ACK/NACK/CQI

MACD PDU

Data Pool

ACK/NACK

CQI

Select Tbsize

Retrans

Que

Select Sch

UE/Que

HSSCCH Info

HSDSCH DataUu

UE1

UE2

Iub

Flow

Control

Entity

Flow

Control

Entity

<=Hsdpa Capa

Alloc Frame

=>Hsdpa Capa

Request Frame

<=Hsdpa Capa

Alloc Frame

The implementation procedure of Node B MAC-hs is as follows:

1 Upon receiving the Frame Protocol (FP) data from lub interface, MAC-hs stores FP

data in the buffer area of each UE.

2 Hsdpa Scheduler selects the UE with data to be transmitted based on the

scheduling algorithm and selects idle HARQ PORCESS of UE through Harq Entity.

3 Hsdpa Scheduler selects appropriate HS -PDSCH air interface transmission format

in the AMC module based on channel quality, terminal capability, available power

and code resources of UE, and then transmits FP data over air interface.

MAC-hs of Node B also implements real-time flow control of downlink HS-DSCH data

frames on lub interface based on the air interface rate and channel quality of UE to

realize sharing of air interface and lub interface bandwidth among multiple UEs.

This document focuses on the description of Hsdpa Scheduler and Flow Control Entity,

that is, the components in grey in Figure 1.

2.1 HSDPA Fast Scheduling

The scheduling procedure at MAC-hs layer: Select the UE with data to be transmitted

based on scheduling algorithm; determine transmission format based on the channel

quality, available code and power resources of UE; transmit downlink data to UE over

HS-SCCH and HS-PDSCH; UE returns data receiving information to MAC-hs of Node B



over HS-DPCCH. If UE returns data receiving failure to MAC-hs, MAC-hs layer of Node

B needs to retransmit the data that originally fails to be received by UE.

The Transmission Time Interval (TTI) of HS-PDSCH is 2 ms as stipulated in the protocol,

which means the scheduling time of MAC-hs layer must be controlled within 2 ms.

Compared with the minimum TTI of R99-10 ms, the scheduling rate of MAC-hs layer

improves by 5 times.

The scheduling algorithms of MAC-hs layer includes the Round Robin (RR), Maximum

Carrier to Interference (MaxC/I) and Proportional Fairness (PF) algorithms. These

algorithms have a great impact on the performance including cell service fairness,

throughput of single UE and cell throughput. As a tradeoff of the RR and MaxC/I

algorithms, the PF algorithm can help achieve large throughput and better service

fairness.

2.2 HSDPA Flow Control

The air interface rate of HSDPA UEs is subject to multiple ever -changing factors such as

UE channel quality, number of UEs to be scheduled in a cell, and code/power resources

of a cell. Therefore, the scheduler necessitates flow control to administrate the incoming

rate of downlink HS-DSCH data frames of each UE to make it consistent with the

outgoing rate of air interface to avoid either excess or insufficient data volume of UE on

Node B side.

On lub interface, the HSDPA scheduler informs RNC to control the transmit rate of

certain UE through the Capacity Allocation Frame (CAF). When certain UE in RNC fails

to receive the CAF within a specified time, RNC will voluntarily send a Cap acity Request

Frame (CRF) to HSDPA scheduler. The HSDPA scheduler then sends a CAF to RNC in

response.

The downlink rate for RNC to transmit data frames to certain UE of Node B scheduler

will not exceed the rate specified in the CAF.

3 Technical Description

3.1 HSDPA Resource Allocation Scheme

HSDPA UEs are scheduled by Node B, while R99 UEs are scheduled by RNC. This

brings about the resource (including code and power) sharing issue among HSDPA and

R99 UEs in the same cell.

The Spreading Factor (SF) of HS-PDSCH is 16, and downlink HSDPA service data is

carried on HS-PDSCH. Maximum number of HS-PDSCH code can be configured to 15

in each cell. RNC informs Node B of the number of HS-PDSCHs that can be used by



certain cell through the Physical Shared Channel Reconfiguration Request signaling of

NBAP.

Node B implements the mechanism of cell power sharing among HSPA and R99 UEs.

As the downlink power control of R99 is implemented in RNC and Node B cannot control

power of R99 UEs, ZTE UMTS Node B HSPA scheduler is designed to support the

function of voluntarily bypassing R99 channel downlink transmit power, that is, the

maximum available power of related HSPA downlink channel shall not exceed the

maximum cell transmit power minus downlink transmit power of R99 channels. The

voluntary bypass function enables cell power sharing among HSPA and R99 UEs.

The procedure for HSPA scheduler to voluntarily bypass R99 channel downlink transmit

power: In each TTI of 2 ms, the cell TSSI reported through RF minus the power

consumed by HSDPA and HSUPA leaves the R99 power at this moment. Then the

available HSPA power in next 2ms TTI can be obtained by subtracting R99 power from

the maximum cell transmit power.

The RNC parameter HSPA Total Downlink Power Allocation Method

(HsdschTotPwrMeth) can be set to RNC Static Assigning Mode(0), RNC Dynamic

Assigning Mode(1), or Node B Assigning Mode(2). The voluntary power bypass of Node

B always takes effect no matter how the parameter HsdschTotPwrMeth is set. This

parameter is set to Node B Assigning Mode(2) by default, that is, Node B is

responsible for controlling power sharing so as to utilize cell transmit power to the

greatest extent.

When HsdschTotPwrMeth is set to RNC Static Assigning Mode(0) or RNC Dynamic

Assigning Mode(1), RNC can exercise control over the maximum available power of

HSPA, that is, the total HSPA power cannot exceed the value specified by RNC. For

details about RNC controlling total HSPA power, see ZTE UMTS Power Control FD.

3.2 HSDPA Scheduling Algorithms

The chief objective of scheduling algorithms is to calculate the relative priority of all UEs

in each TTI of 2ms, prioritize them and schedule those with higher priority first. The

scheduling algorithms implemented by ZTE UMTS Node B include MAX-C/I, RR and PF

algorithms.

MAX C/ I algorithm focuses on the maximum throughput of a cell. RR algorithm gives

equal scheduling opportunity for all UEs. PF algorithm is a tradeoff of MAX C/I and RR

algorithms.

Among these three algorithms, only the PF algorithm takes into account the HSDPA

service attributes.

For class-I/ -B and SRB over HS-DSCH services, the QoS is guaranteed by setting

different Schedule Priority Indicators (SPIs) in RNC.



For Guaranteed Bit Rate (GBR) configured services, the PF algorithm preferentially

schedules UEs with GBR not satisfied.

Note: UEs with retransmitted data must be scheduled preferentially no matter which

scheduling algorithm is adopted.

The principles of above three scheduling algorithms are introduced as follows:

3.2.1 MAX C/I Algorithm

The MAX C/I algorithm only takes into account the channel quality to maximize cell

throughput. The relative priority of MAX C/I algorithm is given by:

RelativePriority = CQI * TBSIZE

The Channel Quality Indicator (CQI) is fed back by HS-DPCCH of UE. The maximum

MAC-hs Transmission Block Size (TBSIZE) of UE is obtained by querying the CQI

mapping table for UE categories provided by TS 25.214 based on current CQI, UE

categories and number of available HS-PDSCH channelization codes.

3.2.2 RR Algorithm

The relative priority of RR algorithm is given by:

RelativePriority = Current Time – Last Time of UE Scheduling

The unit of time in the above equation is TTI 2ms. Current Time: Refers to current

scheduling time;

It is obvious that RR algorithm has the longest scheduling waiting time.

3.2.3 PF Algorithm

The PF algorithm takes into account both the channel quality and history traffic. That is,

the PF algorithm takes into account both cell throughput and user fairness. As a tradeoff

between fairness and cell throughput, the PF algorithm is generally adopted by default.

The relative priority of PF algorithm is given by:

RelativePriority =WeightofSPI Rate (1+HistoryFlux)

The Schedule Priority Indicator (SPI) refers to the UE scheduling priority configured

through NBAP signaling, and ranges between 0 and 15. The SPI is related to the

services used by UE. WeightofSPI refers to the weight obtained through SPI mapping

which is configured through the parameter SPI_0 SPI_1…SPI15. The values of SPI_0,

SPI_1…SPI15 ranges between 1 and 2000, moreover, the value of SPI_n should less

than the values of SPI_n+1 (n=0, 1, 2…14), the reason for this request is that the



mapping relationship between SPI and WeightofSPI should be accordant, otherwise

RNC configured SPI will be different with the real scheduled priority level. The

recommended default value of SPI_0, SPI_1…SPI15 is set to

[10,12,14,17,21,25,30,36,43,52,62,74, 89,107,128,154] by default. It is not suggested to

randomly configure SPI_0 SPI_1…SPI15 not compliant to the request of SPI_n less

than SPI_n+1, because this will result in difference of RNC configured SPI and the real

scheduled priority level, thereby bringing network service priority chaos.

Generally, Rate presents the current instant speed rate. The calculation of Rate is

different for traditional UE, MIMO UE and DC UE, with it’s formula as following:

For traditional UE and MIMO UE single stream:

Rate= 1(CQI_s) TBSIZE(CQI_s) w

For MIMO UE double sreams:

Rate= 2(CQI_1) TBSIZE(CQI_1)+ 2(CQI_2) TBSIZE(CQI_2)w w

For DC-HSDPA UE:

Rate= 1(CQI_n) TBSIZE(CQI_n) w

CQI_s is the CQI reported by traditional UE and MIMO UE under single stream

scheduling. CQI_1 and CQI_2 is the CQI for primary and secondary stream

reported MIMO UE under double stream scheduling. CQI_n is the CQI for related

carrier reported by DC UE. TBSIZE(CQI_s) , TBSIZE(CQI_1) , TBSIZE(CQI_2)

and TBSIZE(CQI_n) are obtained by querying the CQI mapping table for UE

categories provided by TS 25.214 based on current CQI.

1(CQI_s)w and 2(CQI_1)w refer to the weight obtained through CQI mapping which

is configured through the parameter Channel Quality Weight (ConditionWgt). The value of

ConditionWgt ranges between 1 and 6, and is set to 3 by default. The larger the value of

ConditionWgt, the steeper the mapping relation between Weight of CQI and CQI, that is,

the more scheduling chance the UEs with high CQI have. 1w is for single stream

mapping, 2w is for double stream mapping. For detailed mapping table, see Chapter 5.

The history flux of UE is calculated at intervals of 2 ms, and the accumulated newly

transmitted data increases by TBSIZE, as given in the following equation:

For traditional UE:

HistoryFlux(n) = HistoryFlux (n-1) * 0.96 + TBSIZE, where, TBSIZE is a variable

because the data volume scheduled each time varies. n refers to the times of



history scheduling. HistoryFlux(n) refers to the history flux after n times of

scheduling. TBSIZE refers to the TBSIZE of last scheduling.

For MIMO UE:

HistoryFlux(n) = HistoryFlux (n-1) * 0.96 + TBSIZE1+TBSIZ2, where, TBSIZE1

and TBSIZ2 refer to the transmit block size of primary and secondary stream of

MIMO, if primary or secondary stream is not scheduled or retransmitted in this TTI,

TBSIZE1 or TBSIZ2 is 0.

For DC UE:

HistoryFlux(n) = HistoryFlux (n-1) * 0.96 + TBSIZE1+TBSIZ2, where, TBSIZE1

and TBSIZ2 refer to the transmit block size of two carriers, if it’s not scheduled or

retransmitted in this TTI in the related carrier, TBSIZE1 or TBSIZ2 is 0. This is main

idea of the joint scheduling algorithm for DC UE cross two carriers comparing to

traditional UE’s independent scheduling algorithm on single carrier.

If a UE is stream class user, Guaranteed Bit Rate(GBR) is configured by RNC; If a UE is

I/B class user, a minimal GBR is configured by RNC, It’s named Nominal Bit Rate, NBR

value generally is 16K. Both GBR is get by scheduler through Mac-hs Guaranteed Bit

Rate IE of NBAP. In order to discriminate stream class and I/B class, we defined that if

Discard Timer IE of NBAP is configured, scheduler treat the UE as stream class

service; If Discard Timer IE is not configured, scheduler treat the UE as I/B class service.

HSDPA Scheduler first schedule GBR not satisfied stream class UEs, then schedule

GBR not satisfied I/B class UEs, and schedule GBR satisfied UEs and non-GBR UEs

last.

For the GBR not satisfied UEs, they are prioritized and scheduled by the GBR not

satisfied degree.

The non-GBR UEs and GBR satisfied UEs shall be prioritized and scheduled based on

their relative priorities obtained by adopting the PF algorithm.

3.2.4 Summary of Scheduling Algorithms

The MAX C/I algorithm focuses on the maximum cell throughput, but is seldom adopted

in practice.

The PF algorithm is the most widely used and complicated scheduling algorithm, and

also has the best comprehensive effect.

The RR algorithm is rather simple and generally adopted for comparison test with the PF

algorithm.



3.3 HSDPA TFRC Selection Algorithm

When selecting several appropriate UEs for scheduling in each TTI, the scheduler needs

to determine Transport Format and Resource Combination (TFRC) required for each UE,

for example, TBSIZE, HS-PDSCH code information (start code No. and number of

channelization codes), HS-PDSCH power, HS-SCCH code No. and HS-SCCH power.

For UEs with retransmitted data, the TFRC is selected based on the principle of

retaining TBSIZE, number of HS-PDSCH channelization codes and power and HS-

SCCH power.

For UEs requesting transmission of new data, the TFRC is selected starting from UEs

with high relative priority until code or power resources are used up in the cell.

3.3.1 HS-SCCH Code and Power Selection

The selection of HS-SCCH channelization codes is rather simple. An arbitrary code can

be assigned to HS-SCCH of UE unless according to the protocol, the same HS -SCCH

code must be adopted in the event of scheduling in two consecutive TTIs.

Two types of power control algorithms are provided for HS-SCCH. One of them is

constant power algorithm, that is, 1/4 pilot power is adopted constantly, and the other is

CQI-based outer loop power control algorithm. The former is used for testing and the

latter is set by default. The basic principle of the CQI-based outer loop power control

algorithm is as follows:

The power selection of HS-SCCH is given by:

Pdelta/9hs/s NohspdschEsscchNoEPP CPICHHSSCCH

Where,

PCPICH: Refers to the receive power of pilot channel (Unit: dBm).

Es/Nohsscch is constantly 1.2dB.

Γ: refers to the Measurement Power Offset (MeasPwrOffset) configured for

NBAP signaling.

Es/Nohspdsch = -4.5 + CQI dB;

Pdelta refers to the value obtained based on HS-SCCH BLER outer loop

adjustment.

The maximum HS-SCCH power cannot exceed pilot power, and the minimum HS-SCCH

power cannot be less than -16dB of pilot power.



3.3.2 TBSIZE, Number of HS-PDSCH Channelization Codes, Modulation and Power Selection

Two types of HS-DSCH power control algorithms are provided. One of them is the

average power control algorithm, that is, the average available power of all UEs that can

be scheduled in one TTI, and the other is MPO power control algorithm. The former is

generally used for testing and the latter is used by default. The MPO power control

algorithm is described as follows:

The TBSIZE, number of HS-PDSCH channelization codes, modulation mode and power

selection are closely related to CQI, and the target BLER of UE at the time when CQI is

generated is 10%. The CQI reported by various manufacturers, however, are inacc urate

due to the implementation differences or measurement errors, and therefore must be

corrected. Node B adjusts target CQI by using the CQI reported by UE as well as

decoding results, that is, Node B performs outer loop adjustment of CQI reported by UE

so as to minimize impact brought by measurement errors and implementation difference

among different manufacturers.

Basic concept of HSDPA CQI adjustment algorithm: The CQI offset of UE is initialized to

0, MAC-hs TB decoding results are accumulated. The CQI offset of UE increases by

0.01 each time ACK signal is detected, and decreases by 0.09 each time NACK signal is

detected. The adjusted CQI is obtained by adding the CQI reported by UE to the CQI

offset of UE.

The TBSIZE, number of channelization codes, modulation mode and Reference Power

Adjustment can be obtained by querying CQI mapping table for UE categories provided

in TS 25.214 based on adjusted CQI.

CPICHHSPDSCH PP

Where,

PCPICH: Refers to the receive power of pilot channel (Unit: dBm).

Γ: refers to the Measurement Power Offset (MeasPwrOffset) configured for

NBAP signaling.

Reference Power Adjustment obtained after querying the CQI mapping

table for UE categories (unit: dB).

For example, suppose UE category is 8, Γ is set to 6dB and adjusted CQI is 27, we can

obtain the following information by querying the CQI mapping table for UE categories:

TBSIZE: 14411; number of HS-PDSCHs: 10; modulation mode: 16QAM; is -2dB. If

PCPICH is 33 dBm, then HS-PDSCH power is 37 dBm (33 + 6 - 2). If the number of

available HS-PDSCH channelization codes in a cell is not less than 10, available power

is not less than 37 dBm and the UE data volume to be scheduled is larger than 14411,

then TBSIZE is 14411, the number of HS-PDSCH channelization codes is 10,



modulation mode is 16QAM, power is 37dBm and the to-be-scheduled data of UE is

transmitted.

The above example is only an ideal situation. In practice, code, power and data volume

all may fall short of the requirements of CQI mapping table. On the other hand, UEs are

scheduled in a descending order of relative priority in one TTI. We try to guarantee

resources for UEs with high relative priority and use up code and power resources in a

cell when selecting the number of HS-PDSCH channelization codes and power for one

UE.

As shown in the following figure, suppose UE category, Γ configuration and CQI of both

UEs are identical, for example: UE category: 8; Γ: 6dB; adjusted CQI: 27, and to-be-

scheduled data volume is sufficient. The number of HS-PDSCH channelization codes is

15, and available power of the cell is 6 W. Ten HS -PDSCH channelization codes and

5W power are used for scheduling of the first UE, leaving the rest 5 HS-PDSCH

channelization codes and 1W power. In such a case, the available TBSIZE for the

second UE decreases to 4664.

Figure 3-1 Resource allocation during UE scheduling

HS-PDSCH 10

16QAM

TBSIZE 14411

POWER 37Dbm(5W)

HS-PDSCH 15

POWER 6W

HS-PDSCH 5

POWER 1WHS-PDSCH 5

16QAM

TBSIZE 4664

POWER 30Dbm(1W)

The specific processing is described as follows:

The extended CQI mapping table is called CQI mapping extension table. The CQI

mapping extension table contains the number of channelization codes, modulation mode,

TBSIZE and Es/No. For example, suppose the number of channelization codes is 10

and modulation mode is 16QAM, then the corresponding TBSIZE and Es/No table

entries are listed as follows:

TBSIZE Es/No

6554 14.56

7041 15.00

7430 15.36

7840 15.62

8272 15.89

8729 16.19

9047 16.39

9546 16.71

9894 16.94

10440 17.42



11017 17.93

11625 18.46

12048 18.78

12713 19.26

13177 19.60

13904 20.12

14411 20.50

Assume the power meets the requirement during UE scheduling, and Es/No is given by

the following equation:

Es/No = -4.5 + CQI +

If CQI is 27, then the value of Es/No is 20.5. We can obtain TBSIZE, which is 14411, by

querying the CQI mapping extension table based on the number of channelization codes

(10) and modulation mode (16QAM). The power required in this case is 37dBm.

If the power is insufficient, for example, it is only 36 dBm, Es/No ne eds to be adjusted to

(20.5 – 1), that is, 19.5. Suppose the number of channelization codes is 10, we can

obtain TBSIZE and Es/No, which are 12713 and 19.26 respectively, by querying the CQI

mapping extension table. Then we can calculate the HS -PDSCH power through the

following equation:

HS-PDSCH power = 36 – (19.5 (Adjusted Es/No) – 19.26 (Es/No obtained by querying

the table)) = 35.76 dBm

If the number of available channelization codes is insufficient, we can obtain the

corresponding TBSIZE and Es/No in the similar way as above by querying the CQI

mapping extension table based on the number of available channelization codes and UE

Es/No. Then we can calculate HS-PDSCH power based on Es/No.

If the to-be-scheduled data volume of UE is less than the selected TBSIZE, continue

query of the CQI mapping extension table and lower TBSIZE until it is slightly larger than

the to-be-scheduled data volume of UE according to the HS-PDSCH power (Es/No), the

number of channelization codes and the modulation by the above steps. The principle of

querying the CQI mapping extension table: Firstly, based on the fixed modulation and

fixed number of channelization codes, query the CQI mapping extension table through

decreasing the HS-PDSCH power. If it can’t find out suitable TBSIZE, then continue

query of the CQI mapping extension table through decreasing the number of

channelization codes. If it can’t find out suitable TBSIZE through decreasing number of

channelization codes still, at last, continue query of the CQI mapping extension table

through decreasing the modulation, for example 64QAM down to 16QAM, 16QAM down

to QPSK.

Note: If is less than 0 (for example, UE category: 8; CQI > 25), to reach the testing

standard that the download rate is 85% of the air interface rate, that is, BLER is less



than 10%, the HS-PDSCH power will remain unchanged, that is, PCPICH + Γ. And this is

a non-standard solution.

3.4 HSDPA Flow Control Algorithm

The HSDPA scheduler is essentially used to implement the downlink packet transfer

function, that is, to receive UE data over the lub interface and then transmit it through

the air interface.

HS-PDSCH is shared among multiple UEs in a cell, and the total number of UEs as well

as the channel quality of each UE ever change, resulting in constant change of the

transmit rate over every UE air interface. Therefore, a flow control mechanism is

required to control the UE packet receiving rate over the lub interface to realize a

balance between inflow and out flow of packets in scheduler.

3.4.1 Flow Control Implementation Method

Flow control is implemented through Frame Protocol (FP) on user plane over the lub

interface by using downlink HS-DSCH Capacity Request Control Frame and uplink the

HS-DSCH Capacity Allocation Control Frame.

RNC will send a CRF to the Node B, when necessary, to trigger the flow control. Node B

will also send a CAF to RNC, when necessary, to control the UE packet inflow rate. For

specific formats of the CAF and CRF, see TS 25.435 protocol.

Here are three trigger mechanisms available for flow control, namely, the flow control will

be triggered in one of the following cases:

1 The Node B receives a CRF from the RNC.

2 The number of stacked packets on UE side exceeds the maximum or minimum

threshold.

3 The UE flow control timer times out.

Where, in the first mechanism, it is the RNC that initiates the flow control, while in the

rest ones, it is the Node B that initiates the flow control.

After the flow control is triggered, the RNC data receiving rate of UE -Vin is estimated

first based on the air interface rate of UE; then the parameters in relation to the CAF are

calculated according to Vin; finally, a CAF is sent. The flow control procedure is as

follows:



Figure 3-2 Flow control procedure

Flow Control

Trigger Event

Caculate Vin

Get credits interval

repetition period by Vin

Send Capacity

Allocation Frame

End

To avoid frequent trigger of flow control, we set the minimum interval of UE flow control

to 60 ms.

3.4.2 Flow Control Algorithm

As mentioned above, flow control aims at maintaining a balance between UE packet

input and output. Furthermore, flow control also needs to assure there are sufficient

packets during UE scheduling to make full use of the transmitting capability of UE air

interface. In view of these objectives, we design a flow control algorithm as follows:

We regard each packet buffer area of UE as a leaky bucket. As shown in Figure 4, inV

denotes the inflow rate over the lub interface of every UE, corresponding to the RNC

data transmitting rate. outV BufferTimeVoutH *aim is the transmit rate (outflow

rate) of every UE through the Uu interface. The target height of the leaky bucket can be

computed by BufferTimeVoutH *aim . BufferTime indicates buffer time. Generally

the larger the number of to-be-scheduled UEs, the less outflow rate

BufferTimeVoutH *aim .



Figure 3-3 Leaky bucket flow control scheme

PDU BUFFER

RNC

Node B Capacity Allocation

inV ＝ ),,(AIMDIFFERout

HHVF

outV

CURRENTH

AIMH

DIFFERH = AIMH – CURRENTH

DIFFERH

inV

When inV > outV , the bucket height increases; when inV = outV , the bucket height

remains the same; when inV < outV , the bucket height reduces.

Take 80% of aimH as a median, Node B does not voluntarily initiate the flow control

when the inflow rate falls within the upper and lower thresholds of this median. If inV

exceeds the upper threshold that is usually set to 90% of the bucket height, Node B will

initiate the flow control to decrease the inflow rate inV ; if aimH is less than the lower

threshold that is usually set to 70% of the bucket height, Node B will also initiate the flow

control to increase the inflow rate inV .

Note: The BufferTime of UE leaky bucket is constant, and at present is set to 150 ms by

taking into account various factors. HSDPA scheduler offers a buffer area with constant

size of less than 150M bit for sharing among all UEs. Assume BufferTime is 150 ms, the

buffer area far exceeds current air interface capability requirement: 1000M bit/s.

All UEs share the air interface capability in a cell, so whether there are one or several

UEs in a cell does not make much of a difference for the required buffer area. The value

of BufferTime is constant, so the higher the air interface rate of UE, the more the buffer

data packets used by UE or vice versa.

Otherwise, we employ two aspects processing in HSDPA packet scheduling to

cooperate HSDPA flow control algorithm.

A. According to characteristics of different services, RNC will configure and sending two

parameters to Node B: DISCARDTIMERPRE and DISCARDTIMER, Node B will employ

these two parameters in different services scheduling process. When data packet can’t



be scheduled by Node B in the specific UE ’s data buffer in the predefined time duration,

these data packet will be discarded to ease the congestion in the data buffer of Node B.

We will not configure DISCARDTIMER parameter for services which need accurate data

transmission, such as Interactive and Background class services. But for services which

have high time-delay requirement such as Streaming and Conversation(VoIP over

HSDPA) class, we can configure DISCARDTIMER parameter selectively for them to

discard data packet that can’t be scheduled in time. And the DISCARDTIMER time

duration can be different, for example, 4s for Streaming class and 60-80ms for

Conversation class with higher time-delay QOS requirement.

B. Introducing sliding window management and T1 timer mechanism to stop related

MAC-hs PDU retransmission which couldn’t be accepted by UE even with multiple

retransmission.

Node B sliding window management is a algorithm for MAC-hs PDU transmission

management in scheduling process. And the purpose is to avoid UE receiving MAC-hs

PDU with unclear TSN. At the same time, Node B scheduling algorithm can employ T1

timer to stop MAC-hs PDU retransmission which couldn’t be accepted by UE even with

multiple retransmission.

T1 timer and window size parameter MACHSWINSIZE of MAC-hs transmission and

receiving side can be configured by RNC OMM, and delivered to Node B and UE

through signaling respectively.

Sliding window management is to wait MAC-hs PDU at lower edge of the the window

received accurately, or until retransmission times and T1 timer expiring, the window will

slide forward. Assuming TrsWindow_LowerEdge is the lower edge of the the window,

TrsWindow_LowerEdge represents minimum TSN of the MAC-hs PDU which is waiting

for ACK response. If receiving ACK response of this MAC-hs PDU successfully or

discarding this MAC-hs PDU because of retransmission times and T1 timer expiring,

TrsWindow_LowerEdge will slide to the next minimum TSN of the MAC-hs PDU which is

waiting for ACK response. MAC-hs PDU with smaller TSN than TrsWindow_LowerEdge

will not be retransmitted, and MAC-hs PDU with bigger TSN than

TrsWindow_LowerEdge + MACHSWINSIZE also can’t be transmitted(waiting to be

transmitted) .

For example, MACHSWINSIZE is 12,TrsWindow_LowerEdge is 0,as the following figure,

TSN=12,13,14 MAC-hs PDU outside of window are not allowed to be transmitted, if

receiving NACK response for TSN=0 MAC-hs PDU and discarded because of

retransmission times and T1 timer expiring, receiving ACK response for TSN=1

MAC-hs PDU and NACK for TSN=2 MAC-hs PD, TrsWindow_LowerEdge will silde to 2,

TSN=12,13 MAC-hs PDU now are allowed to be transmitted, TSN=14 MAC-hs PDU is

still waiting to be transmitted, TSN<2 MAC-hs PDU is not allowed retransmitted again.

The default configuration for MACHSWINSIZE is 16, and specially for MIMO dual stream

transmission service, MACHSWINSIZE is configured 32 defaultly.



1 2 3 4 5 6 7 8 9 10 11 12

TRANSMIT_WINDOW_SIZE

TRANSMIT_WINDOW_SIZE

0 13 14

Node B can employ T1 timer to stop ret ransmission of related MAC-hs PDU.

Firstly, at the UE side, if no timer T1 is active, the timer T1 shall be started when a MAC-

hs PDU with TSN > next_expected_TSN is correctly received, that is to say, UE will start

T1 based on successfully receiving next TSN MAC-hs PDU. But Node B does not know

accurately whether UE received next TSN MAC-hs PDU or not, so Node B will start T1

after receiving ACK response for next TSN MAC-hs PDU from UE. ACK response delay

or lost will result in Node B starting T1 later than UE, so Node B will properly postpone

discarding related PDU. On mechanism of retransmission priority and retransmission

number limitation, Node B will guarantee to discard specific MAC-hs PDU in time which

can’t be transmitted and will not block next TSN PDU transmission significantly.

T1 timer range is 【10ms-400ms,…】and the default configuration is 50ms at RNC

OMM.

3.5 Measure of HS-DSCH Required Power

According to TS25.433, there are three types of common measure: Transmitted carrier

power of all codes not used for HS -PDSCH or HS-SCCH transmission, HS-DSCH

Required Power, HS-DSCH Provided Bit Rate. Among these common measures, HS-

DSCH Required Power is a key refrence parameter of RNC admission control. Here is

detailed description of HS-DSCH Required Power.

According to TS25.433, HS-DSCH Required Power Value indicates the minimum

necessary power for a given priority class to meet the GBR for all UEs belonging to this

priority class. It is expressed in thousandths of the cell max transmission power.

The basic principle is to pre-calculate required power for GBR rate with 100ms time

duration according to recent historical power and throughput rate.

Node B can schedule multiple packet data QUEs for every UE. For one scheduled

packet data QUE, the required power is the sum of HS-SCCH required power and

HS-DSCH required power. HS-SCCH required power can be calculated by HS-

SCCH reserved power/QUE number. And HS-DSCH required power calculation

process as following:

1. Get CQI according to GBR:

GbrTbSize = GBR / 500,check TS25.214 CQI table, get GbrCqi according to

GbrTbSize.



2. Statistics of scheduled packet data QUE every 100ms:

HS-DSCH accumulated power: AccuDschPwr

Accumulated transmitted BITNumber: AccuSchBitNum

3. According the actual scheduling, calculate the actual corresponding CQI:

SchTbSize = AccuSchBitNum / SchNum, thereinto, SchNum is the scheduled

times in 100ms, check TS25.214 CQI table, get SchCqi according to SchTbSize.

4. Calculate HS-DSCH required power.

The actual average HS-DSCH power is: SchDschPwr = AccuDschPwer /

SchNum

HS-DSCH required power = SchDschPwr - (SchCqi - GbrCqi)

At the same time, pay attention that HS-DSCH required power can’t higher than

pilot power+MPO, and can’t lower than pilot power -10dB。

5. Filtering HS-DSCH required power, then report to RNC:

The current reported HS-DSCH required power = last time reported HS-DSCH

required power * (1- gdfMeasFilterCoeff) + current calculated HS-DSCH

required power * gdfMeasFilterCoeff, thereinto the filtering factor

gdfMeasFilterCoeff = 0.015625.

3.6 Impact of HSPA+ on Scheduler

3.6.1 Impact of introducing 64QAM Modulation Technology on Scheduler

After introducing the HSDPA and HSUPA technologies respectively in R5 and R6

protocol versions, 3GPP organization is now introducing new technologies in R7 and

subsequent versions to enhance HSPA performance. After the HSDPA technology is

introduced to WCDMA, the MAC-ehs layer is added to both Node B and UE and

compatible with all functions of original MAC-hs layer. For the HSDPA scheduler of Node

B, R7 is primarily characterized by two technologies: 64QAM and improved L2.

Introducing of HSPA+ technology does not have any impact on the HSDPA scheduling

and flow control algorithms. But it has certain impact on HS-SCCH/HS-PDSCH code

selection, and packetization of MAC-ehs PDU.

After the 64QAM modulation technology is adopted, 64QAM improves the modulation

efficiency by 50%, and accordingly, the peak rate of single UE increases by 50% and

reaches 21.6 Mbps compared with HS -DSCH of R5.



After 64QAM technology is introduced, the major impact on HSDPA scheduler lies in the

change of modulation mode information carried on HS-SCCH and HS-PDSCH code-set

information.

The modulation mode information and HS-PDSCH code-set information carried on HS-

SCCH are defined as follows according to R5 TS25.212.

xccs,1, xccs,2, xccs,3 = min(P-1,15-P) P refers to the number of HS-PDSCH

channelization codes.

xccs,4, xccs,5, xccs,6, xccs,7 = |O-1-P/8 *15| O refers to the start code No. of HS-

PDSCH.

QAMif

QPSKifxms 161

01,

After 64QAM modulation mode is introduced, if xms,1 is 0, the modulation mode is QPSK;

if xms,1 is 1, the modulation mode is 16QAM or 64QAM.

otherwise

QPSKifxms 1

01,

To further differentiate between 16QAM and 64QAM modulation modes, the lowest BIT

xccs,7 in HS-PDSCH code information bits (xccs,4, xccs,5, xccs,6, xccs,7) is used. Specifically, if

xccs,7 is 0, the modulation mode is 16QAM; if xccs,7 is 1, the modulation mode is 64QAM.

QAMif

QAMifxccs 641

1607,

The lowest BIT position (xccs,7 in R5) of |O-1-P/8 *15| is subject to the even/odd

position of HS-SCCH code demodulated by UE. If HS-SCCH code position is 1, or 3,

xccs,7 is 1; if HS-SCCH code position is 2, or 4, xccs,7 is 0; That is, the following equation

must be met:

|O-1-P/8 *15| mod 2 = (HS-SCCH number) mod 2

If 16QAM or 64QAM modulation mode is selected, then the available HS-PDSCH start

code Nos. are listed as follows:

Table 3-1 HS-PDSCH start code Nos

Odd position of HS-SCCH code

Even position of HS-SCCH code

P >= 8 1 3 5 7 2 4 6

P < 8 2 4 6 8 10 12 14 1 3 5 7 9 11 13 15



To break the downlink data transmission rate bottleneck caused by fixed AM RLC PDU

length and RLC transmit window size defined in R6 and earlier versions, the improved

L2 technology enables the RLC layer of RNC to support variable RLC PDU lengths. The

corresponding MAC-ehs SDU length changes, and the length field L in MAC-ehs PDU is

used to indicate MAC-ehs SDU length.

The multiplexing of logical channel data shifts from RNC to Node B, that is, several

dedicated logical channels are multiplexed into one MAC-d stream and distinguished

through Logical Channel IDs (LCH-IDs) in MAC-ehs PDU. When MAC-ehs is used, LCH-

ID multiplexing replaces C/T multiplexing in MAC-d. The HSDPA scheduler receives and

stores LCH-ID field from HS-DSCH DATA FRAME TYPE2, and forwards this field

through MAC-ehs PDU. For the frame format of HS-DSCH DATA FRAME TYPE2, see

TS25.435.

MAC-ehs SDUs are of variable lengths. To avoid failure in transmitting a complete MAC-

ehs SDU because the length of MAC-ehs SDU exceeds that of MAC-ehs PDU, MAC-

ehs PDU has a Segmentation Indication (SI) field added to support segmented

transmission of MAC-ehs SDU.

For detailed format of MAC-ehs PDU, see TS25.321.

When both Node B and UE support improved L2, RNC determines the type of signaling

and service, that is, fixed or flexible PDU size, based on the settings of the parameter

RlcSizeSuptType. This parameter is a switch used to indicate RLC PDU size type

configured for signaling and services. It has two values:

1: Signalling Fixed Mode, Service Flexible Mode

2: All Flexible Mode

The RLC information reassignment is required when a UE takes handover between cells

that support flexible size and those that do not or when transport channel type switches

between HS-DSCH and DCH or FACH. In the case of LI s ize inconsistency after

handover or switching, RESET or MRW procedure must be initiated to re-establish RLC.

To avoid signaling RB RLC reestablishment, fixed PDU Size is recommended for

signaling RB, RLC PDU size is set to 144bits and LI size is constantly 7bits. After new

technologies including 64QAM are used, flexible PDU size must be used for service RB,

so RlcSizeSuptType is set to 1 by default.

The maximum length of Data field in RLC PDU is 1504 bytes (including lengths of SN

and LI) as stipulated in the protocol. The actual maximum available SDU size is

configured by RNC through the parameter CMaxPDUSize and NonCMaxPDUSize.

CMaxPDUSize is configured for conversational service and NonCMaxPDUSize is

configured for non conversational service. In view of protocol header overhead at bottom

layer (FP layer, IP/UDP layer and so on) during data transmission, NonCMaxPDUSize is

set to 600 bytes (4800 bits) by default to avoid decrease of efficiency due to

segmentation at transmission layer.



3.6.2 Impact of introducing MIMO on Scheduler

Impact of introducing MIMO in R7 on scheduler includes the following aspects: PF

scheduling algorithm needs to be modified according to MIMO UE; We need adding

MIMO single/dual stream selecting algorithm in front of scheduling priority calculation.

MIMO single/dual stream selecting algorithm is to preliminarily choose single or dual

stream transmission for the MIMO UE in the current TTI. Impact on TFRC selection

algorithm; And VAM technology.

Let’s expound these impact individually.

1．PF scheduling algorithm needs to be modified according to MIMO UE

Please refer to section 3.2.3 for details related to MIMO UE.

2. MIMO single/dual stream selecting algorithm

UE decides preferred precoding control indication (PCI) vector matrix pref

2

pref

1 , ww

combining with CQI and transmits to Node B through uplink HS-DPCCH signaling.

Based on PCI/CQI compositive report, Node B packet scheduling module will decide

and send to UE with single or dual stream transmission mode, TBSize, modulation mode

in next TTI. At the same time, Node B will inform UE precoding weight w2 which is put in

precoding weight indication bit in the first part of downlink HS-SCCH subframe.

Protocol defined PCI and CQI compositive coding format when UE is configured MIMO

working mode. And UE must support two types of CQI report: type A or type B CQI

report. Type A or Type B CQI report with CQI range （0~30） for single stream

transmission, Type A CQI report with CQI range（0~255）for dual stream transmission.

Type A CQI report is a CQI report format which UE decides transmission block

number(1 block or 2 blocks) according to current channel condition. When 1

transmission block is selected by UE, the preferred primary precoding vector(PCI) from

HS-DPCCH will be used in Node B to precode the primary transmission block. When 2

transmission blocks are selected by UE, the preferred primary precoding vector and

orthogonal precoding vector (PCI) will be used in Node B to precode the primary and

secondary transmission blocks. Type A CQI report will include 1 or 2 transmission format

according to the transmission block number.

Type B CQI report is a CQI report format which UE decides single transmission block

according to current channel condition. The preferred primary precoding vector(PCI)

from HS-DPCCH will be used in Node B to precode the primary transmission block, and

there is no secondary transmission block.

For these two different types of CQI report－type A and type B, UE will achieve different

report rate which is configured by RAN through the following method:



Report rate for type A CQI report is N_cqi_typeA/M_cqi, rest（M_cqi-N_cqi_typeA） will

be used for type B CQI report. MIMO CQI report rate that can be used is the same as

SISO, which is the function of CQI feedback duration k and CQI repeating factor

N_cqi_transmit.

According to TS25.214v7.9.0, when the following formula is tenable:

_cqi_typeA _cqimod76802565

NMk

chipchipmCFN

with

)2( mskk , UE will send type A CQI report, otherwise send type B CQI report.

When MIMO Activation Indicator in HS-DSCH FDD Information signaling is sent by RNC,

Node B will feed back MIMO N/M Ratio（N_cqi_typeA/M_cqi）to RNC in HS-DSCH

Information Response signaling. RNC will send MIMO N/M Ratio to UE through RRC

signaling. And UE will send type A and type B CQI report according to MIMO N/M Ratio,

Node B will also receive type A and type B CQI report according to MIMO N/M Ratio.

MIMO N/M Ratio value can be:1/2, 2/3, 3/4, 4/5, 5/6, 6/7, 7/8, 8/9, 9/10, 1/1. Now we

only consider to use MIMO N/M Ratio with 1/2 and 1,and it’s default configuration is 1,

that is to say UE will always report Type A CQI report.

MIMO single/dual stream selecting algorithm flow description is as following:

1) After outer loop process of the UE reported CQI, Node B will judge UE CQI report is

Type A or Type B. If it is Type A, go to step 2, if it is Type B, go to step 3.

2) For Type A CQI, if it’s single stream CQI, Node B will employ single stream

scheduling in the current TTI; If it’s dual stream CQI, Node B will employ dual

streams scheduling in the current TTI;

3) For Type B CQI, Node B will check the recent historical type A CQI report is single

stream or dual streams before the current TTI. If it ’s single stream, Node B will

employ single stream scheduling for UE in current TTI based on Type B CQI report.

If it’s dual stream, comparing the current TTI single stream TBSize and sum of

recent historical dual streams TBSize, if current TTI single stream TBSize is bigger,

Node B will employ single stream scheduling for UE in current TTI based on Type B

CQI report, otherwise Node B will employ dual streams scheduling based on Type

A CQI report;

4) Under small data buffer conditions, dual streams process will change into single

stream process.

3. Impact on TFRC selection algorithm

TFRC selection algorithm for MIMO UE single stream scheduling is the same as

traditional UE in section 3.3. But there is some difference for MIMO UE double stream

scheduling in TFRC selection algorithm:

1） Firstly, modify the CQI of MIMO primary and secondary stream individually with

the same principle as in section 3.3.



2） According to modified CQI of primary and secondary stream, without

consideration of code and power resource, to select the TBSIZE of two streams,

then calculate the needed power, and calculate power per code channel and

code rate.

3） According to the cell rest power, to calculate UE code channel number that can

be used based on the principle of no change to power of per code channel

4） According to code channel number that can be used, to select the TBSIZE

based on no change to code rate

5） Finally to modify TBSIZE according to packet data volume waiting for

scheduling.

4. VAM technology

Primary and secondary pilot configuration with VAM (Virtual Antenna Mapping)

technology will be deployed when MIMO UEs and traditional non-MIMO UEs are

supported in mixed networking. VAM technology is mainly used for power balance

between two PAs and load balance between two transmission channels. And for MIMO

UE, under single stream scheduling condition, PCI codebook restrictions of

2

1

2

1pref

2

jjw on UE side is needed to assure two PAs power balance. In

order to reduce the affection of secondary pilot to traditional non-MIMO UEs, secondary

pilot power can be configured as half of primary pilot, but few MIMO commercial

terminals support this kind of configuration.

3.6.3 Impact of introducing DC-HSDPA on Scheduler

After introducing DC-HSDPA in R8, new joint PF scheduling algorithm will be used for

DC UE, and scheduling relative priority formula for DC UE need to be adjusted

accordingly. Please refer to section 3.2.3 for details related to DC UE.

The queuing and scheduling rule according UE relative scheduling priority:

On carrier1, single carrier UEs and DC UEs will be queued and scheduled according to

described relative priority;

In the same way, on carrier2, single carrier UEs and DC UEs will be queued and

scheduled according to described relative priority.

3.7 Dynamic power sharing in Multi-carrier

Dynamic power sharing in multi-carrier function will be supported with upgrading

software of Node B. The advantage of this technique is to improve utilization rate of

output power in multi carriers and improve quality of user service, increase downlink

capacity, and decrease TCO(Total Cost of Owernship).



Traditional power sharing in multi-carrier is static, power of each carrier is the same. For

example, show as in figure 5. In practical field application, power of multi-carrier can not

be totally used. Load threshold of R99 is low, power of R99 is surplus most of the time.

After introducing HSDPA, HSDPA scheduler can adopt strategy of dynamic power

allocation, total output power of cell can keep a smooth level, and interference of

downlink is more stable, so load threshold of HSDPA can be higher, show as Figure 3-4.

Figure 3-4 Static power sharing in Multi-carrier

Pmax

Pmax/2carrier2carrier1R99

HSDPA+R99

Figure 3-5 Downlink power after introducing HSDPA

HS-DSCH (rate controlled)

DCH (power controlled)

Common channels

t

power

To

tal

cell

po

wer

To

tal

cell

po

wer

Common channels

DCH (power controlled)

Unused power

power

t

HS-DSCH with dynamic power allocationPower usage with DCH

R99 R5

For example, carrier1 is R99 service carrier, carrier2 is R99+HSDPA service carrier,

total output power of equipment is Pmax, Pmax-carrier1 and Pmax-carrier2 are max

power for each carrier . Pth is the power which can be shared between carriers, Pth=k*

min（Pmax-carrier1,Pmax-carrier2）, k is sharing power ratio which can be configured

in OMM , such as 10% or 20%. Pmax and K corresponds to parameters of dynamic

power sharing in multi-carrier Transmission Power and Power-sharing Ratio in chapter

4.2。



Then take an actual instance for example, each carrier of UMTS Node B is 20W. At time

T, output power of carrier1 is 13W, output power of carrier2 is 17W, but carrier2 need

more 4W for HSDPA service, and based on dynamic power sharing in multi-carrier, 4W

of carrier1 can share with carrier2 to satisfy HSDPA service requirement, and capacity of

network is also enhanced.

4 Parameter Description

4.1 HSDPA Scheduling Algorithm Parameters

Table 4-1 Parameter settings in OMMB

Abbreviated name Parameter name

ConditionWgt Channel Quality Weight

SPI_0 SPI_1…SPI_15 SPI Factor

Table 4-2 Parameter settings in OMMR


HsdschTotPwrMeth HSPA Total Downlink Power Allocation Method

MeasPwrOffset HS-PDSCH Measurement Power Offset

RlcSizeSuptType Support Type of RLC Flexible PDU Size Format

NONCMAXPDUSIZE Non-Conversational service Maximum MAC-d PDU Size Extended(byte)

CMAXPDUSIZE Conversational service Maximum MAC-d PDU Size Extended(byte)

4.1.1 Channel Quality Weight

OMM Path

View->Configuration Management ->NodeB NE->Base Station Radio Resource

Management ->WCDMA Radio Resource Management->Baseband Resource Pool-

>Local Cell ->HSDPA parameter-> Channel Quality Weight

Parameter Configuration

This parameter indicates channel quality weight. The value of this parameter is [1 6]. Its

default value is 3.



4.1.2 SPI_0 SPI_1...SPI_15

OMM Path


Management ->WCDMA Radio Resource Management->Baseband Resource Pool-

>Local Cell ->HSDPA parameter-> SPI_0 SPI_1…SPI_15


This parameter indicates weight of Que SPI. The values of SPI_0, SPI_1…SPI15 may

be modified. The values of SPI_0, SPI_1…SPI15 ranges between 1 and 2000, moreover,

the value of SPI_n should less than the values of SPI_n+1 (n=0, 1, 2…14).

The default value of this parameter is [10, 12, 14, 17, 21, 25, 30, 36, 43, 52, 62, 74, 89,

107, 128, 154].

4.1.3 HS-PDSCH Total Downlink Power Allocation Method

OMM Path

View->Configuration Management->RNC NE->RNC Radio Resource Management-

>Modify Advanced Parameter ->HSPA Configuration Information


This parameter is an internal parameter of RNC. It indicates the method for allocation of

total HS-PDSCH power. Three allocation methods are supported:

0：RNC Static Assigning Mode

1：RNC Dynamic Assigning Mode

2：Node B Assigning Mode

By default, this parameter is set to 2: Node B Assigning Mode.

4.1.4 HS-PDSCH Measurement Power Offset

OMM Path

View->Configuration Management->RNC NE->Rnc Radio Resource Management-

>UtranCell->UtranCellXXX-> Modify Advanced Parameter ->Hspa Configuration

Information In A Cell




This parameter indicates the assumed HS-PDSCH power offset relative to P-CPICH/S-

CPICH power used for CQI measurement. The range of this parameter is [-6, 13] dB by

step 0.5dB.

Its default value is 6dB.

4.1.5 Support Type of RLC Flexible PDU Size Format

OMM Path


>Modify Advanced Parameter ->HSPA Configuration Information


This parameter is a switch used to indicate RLC PDU size type configured for signaling

and services. It has 2 values:

1: Signalling Fixed Mode, Service Flexible Mode

2: All Flexible Mode

By default, this parameter is set to 1: Signalling Fixed Mode, Service Flexible Mode

4.1.6 Non-Conversational service Maximum MAC-d PDU Size Extended

OMM Path


>Advanced Parameter Manage->HSPA Configuration Information -> Non-

Conversational service Maximum MAC-d PDU Size Extended(byte)


This parameter indicates the Non-Conversational service maximum size in octets of the

MAC level PDU when an extended MAC level PDU size is required. According to the

protocol, the maximum length of the Data field in an RLC PDU is 1504 bytes (12032 bits,

including the length of the SN and the length of the LI). Considering the protocol header

overheads of the bottom layer (FP layer, IP/UDP layer, etc.) during data transmission, the

RNC sets the Non-Conversational service maximum MAC-D PDU size to 600 bytes by

default so as to avoid low transmission efficiency caused by segmentation in the

transport layer.

4.1.7 Conversational service Maximum MAC-d PDU Size Extended

OMM Path



View->Configuration Management->RNC NE->RNC Radio Resource Management->

Modify Advanced Parameter -> HSPA Configuration Information -> Conversational

service Maximum MAC-d PDU Size Extended


This parameter indicates the Conversational service maximum size in octets of the MAC

level PDU when an extended MAC level PDU size is required. The RNC sets the

Conversational service maximum MAC-D PDU size to 65 bytes by default.

4.2 Dynamic Power Sharing in Multi-carrier Parameters

Table 4-3 OMMB configuration parameters filed


Transmission Power Local Cell Transmission Power

Power-sharing Ratio Power-sharing Ratio

4.2.1 Transmission Power

OMM path


Management -> WCDMA Radio Resource Management->Sector Management->Sector-

> Local Cell ->Transmission Power


This parameter is the maximum transmission power of local cell, with range of [0,200].

4.2.2 Power-sharing Ratio

OMM path

View->Configuration Management ->NodeB NE->Base Station Equipment Resource

Management ->Power-sharing Ratio


This parameter is configured for power-sharing ration in dynamic power-sharing in multi-

carriers, with the range of [1, 50%].



5 Weight Mapping Table

Table 5-1 CQI weight mapping table

Wgt1 Wgt2 Wgt3 Wgt4 Wgt5 Wgt6

Single stream CQI w1

1 1 1 1 1 1 1

2 1 2 4 1 1 1

3 1 3 9 1 1 1

4 1 4 16 1 1 1

5 1 5 25 1 1 1

6 1 6 36 2 1 1

7 1 7 49 3 2 2

8 1 8 64 5 4 3

9 1 9 81 7 7 6

10 1 10 100 10 10 10

11 1 11 121 13 15 16

12 1 12 144 17 21 25

13 1 13 169 22 29 37

14 1 14 196 27 38 54

15 1 15 225 34 51 76

16 1 16 256 41 66 105

17 1 17 289 49 84 142

18 1 18 324 58 105 189

19 1 19 361 69 130 248

20 1 20 400 80 160 320

21 1 21 441 93 194 408

22 1 22 484 106 234 515

23 1 23 529 122 280 644

24 1 24 576 138 332 796

25 1 25 625 156 391 977

26 1 26 676 176 457 1188

27 1 27 729 197 531 1435

28 1 28 784 220 615 1721

29 1 29 841 244 707 2051

30 1 30 900 270 810 2430

Double stream CQI w2



Wgt1 Wgt2 Wgt3 Wgt4 Wgt5 Wgt6

0 1 21 426 84 197 510

1 1 21 426 84 197 510

2 1 22 479 124 254 769

3 1 26 620 142 374 780

4 1 27 692 178 392 786

5 1 27 732 203 524 1380

6 1 28 838 230 692 2068

7 1 30 880 280 823 2520

8 1 30 925 282 833 2534

9 1 32 958 330 989 2900

10 1 34 1042 386 1199 3786

11 1 35 1197 462 1579 5346

12 1 36 1278 510 1755 6204

13 1 37 1364 561 1917 7172

14 1 38 1449 599 2081 8162

6 Glossary

3GPP 3rd Generation Partnership Project

A

ACK ACKnowledgement

AMC Adaptive Modulation and Coding

ARQ Automatic Repeat Request

B

BLER Block Error Rate

C

CPICH Common Pilot Channel

CQI Channel Quality Indicator

C/I Carrier / Interference

D

DCH Dedicated Channel



DL Downlink (Forward link)

F

FACH Forward Access Channel

G

GBR Guaranteed bit rate

H

HARQ Hybrid Automatic Retransmission Request

HS-DPCCH High Speed Dedicated Physical Control Channel

HS-DSCH High Speed Downlink Shared Channel

HS-PDSCH High Speed Physical Downlink Shared Channel

HS-SCCH High Speed Shared Control Channel

HSDPA High Speed Downlink Packet Access

HSPA High Speed Packet Access

HSPA + High Speed Packet Access Plus

HSUPA High Speed Uplink Packet Access

K

Kbps kilo-bits per second

M

MAC Media Access Control

Mbps Mega-bits per second

MPO Measure Power Offset

N

NACK Negative ACKnowledgement

NBR Nominal Bit Rate

O



OMM Operations & Maintenance Management Center

OMMB OMM of the Node B

OMMR OMM of the RNC

P

P-CPICH Primary Common Pilot Channel

PDU Protocol Data Unit

Q

QAM Quadrature Amplitude Modulation

QoS Quality of Service

QPSK Quaternary Phase Shift Keying

R

RAB Radio Access Bearer

RLC Radio Link Control

RNC Radio Network Controller

RR Round Robin

RRC Radio Resource Control

S

S-CPICH Secondary Common Pilot Channel

SDU Service Data Unit

T

TFC Transport Format Combination

TFRC Transport Format and Resource Combination

TTI Transmission Time Interval

U

UE User Equipment



UMTS Universal Mobile Telecommunications System

UTRAN Universal Terrestrial Radio Access Network

W

WCDMA Wideband Code Division Multiple Access

ZTE UMTS HSDPA Packet Scheduling Feature Guide

Documents

hsdpa scheduling algorithms

zte corporation

hsdpa fast scheduling

hsdpa flow control algorithm

hsdpa packet scheduling

hsdpa tfrc selection

proprietary information

improvement of zte products